E-commerce is changing its grammar. For two decades, brand visibility was built around keywords, SEO and product pages. Today, with the rise of conversational interfaces, another logic is gradually taking over: that of context.
To talk about it today we welcome the very media-oriented Roxanne Laigle
According to Roxane Laigle, CEO of LEMROCK. “Tomorrow, we will no longer buy on sites, but in conversations. » Behind this formula, the emergence of models like OpenAI or Google with Gemini not only redefines the search for information, but is already reconfiguring the first points of contact between brands and their customers.
From keyword to context, a structural mutation
In the old model, the purchase intention was explicit. The user formulated a request such as “sunglasses”, “cordless vacuum cleaner”, and the market was organized around this request. The entire economy of SEO and retail media was based on this ability to capture a formulated intention.
In conversational interfaces, this logic gradually disappears. The user describes a situation, a project, a constraint. He talks about vacations, a specific need, a use. The purchasing intention becomes implicit, sometimes even diffuse.
“We no longer fight to get back on a keyword, but to be recommended in a context,” explains Roxane Laigle.
This change is profoundly shaping the way products are discovered, and imposing new information processing models capable of interpreting weak signals and producing relevant recommendations in real time.
LEMROCK, an infrastructure between brands and AI
It is in this area that LEMROCK is positioning itself, founded in 2025, the startup is developing a platform allowing e-retailers to connect their catalogs to language models and optimize their presence in these new environments.
“Today, there is a missing layer to connect brands to conversational AI simply, securely and at scale. This is precisely what we are building. »
This “layer” meets several constraints. On the one hand, the models are based on specific technical protocols, still in the process of being standardized. On the other hand, brands do not wish to directly expose their strategic data (enriched catalogs, margins, prices) to third parties.
In this context, LEMROCK positions itself as an intermediary capable of orchestrating the presence of brands on several models, while maintaining a level of control and confidentiality.
A fragmented market, far from the Google model
Unlike search, dominated by a single player, the conversational AI market appears more fragmented. Several models coexist, with differentiated uses, specific audiences and varying levels of engagement. “Unlike Google, the LLM market will be fragmented. For a brand, it will be essential to be present in several interfaces, general or specialized. »
This fragmentation justifies its agnostic approach, capable of distributing catalogs across different models and agents, without depending on a single actor.
SEO, GEO, agentic commerce: a new value chain
The emergence of LLMs does not eliminate existing logics, but superimposes them. Roxane Laigle thus distinguishes several levels of intervention.
GEO, first, aims to influence how a brand appears in models’ natural responses. It extends, in another form, the logic of SEO.
Agentic commerce, then, is part of a different logic. It does not seek to modify the response, but to enrich it with a transactional layer, by connecting an intention to a product.
“We are not changing the response of the models. We start from this answer to add a layer of commerce. » This distinction outlines a new value chain, where visibility, recommendation and transaction are shared between several players.
An adoption already underway
Although the market is in its infancy, LEMROCK boasts already tangible traction and claims to be deployed in significant volumes of conversations and to work with more than a hundred brands, including some major brands.
“We are no longer in tests. We have paying clients, committed over several months, with significant budgets. »
The advanced figures reflect a notable difference with traditional channels. The interaction rates observed would be significantly higher, driven by the personalization and granularity of conversational exchanges.
An economic model aligned with performance
On an economic level, LEMROCK adopts a model based on performance. Compensation is correlated to results generated for customers, whether through interactions or sales. “Our model is simple: the more money our customers make, the more we make. »
This positioning aims to facilitate adoption, limiting the initial risk for brands. It also brings LEMROCK closer to historical adtech models, while adapting them to a conversational environment.
A speed race against the big platforms
The question of competition remains open. If LEMROCK claims a lead in this segment, the company recognizes that new players (startups, e-commerce platforms, advertising specialists) will quickly position themselves.
In the background, large technology platforms constitute a determining variable. But for Roxane Laigle, their role does not call into question the existence of intermediaries. “There is a structural need for an intermediate layer, for reasons of control, interoperability and data management. » A strategic bet, while players like Meta Platforms are strengthening their ambitions in commerce integrated with AI.
Towards a new commercial standard?
With a fundraising of 6 million euros and an international ambition, LEMROCK intends to accelerate, particularly on the American market. The stated objective goes beyond just commercial growth: it is about defining a standard. “Our ambition is to build the benchmark infrastructure for agentic commerce. »
It remains to be seen whether this standard will emerge around independent players or be absorbed by the platforms themselves. One thing, however, seems certain: as conversational interfaces become established, the battle for commercial visibility is changing in nature.
And in this new landscape, it is no longer the keywords that will make the difference, but the ability to exist in a conversation.
